Ascertaining the cause/s of differences in performance between units of the public administration is at least as important as quantifying the very level of individual performance. We identify two causes: inefficiency and congestion. Inefficiency might be due to the fact that inputs are of relatively low quality (e.g, temporary workers) and/or, for a given quality, to the fact that inputs do not have incentives to exert their maximum level of effort. Congestion arises when the number of new cases is above the number that can be solved when full efficiency is achieved. This decomposition is applied to the universe of Catalonian first instance courts for the period 2005-13 applying a fixed-effect panel stochastic frontier model (Wang and Ho, 2010). In this particular case, we conclude that poor performance is not due to inefficiency on the courts' side, but to an increase in the litigation demand, that is, to congestion, while inefficiency tends to decrease along time and is correlated with the presence of temporary judges.